A recent multicenter diagnostic study has prospectively validated a prediction model using eight nonimaging parameters to diagnose acute pancreatitis in patients visiting the emergency department with elevated serum lipase levels. 

In the study, published in JAMA Network Open, researchers enrolled 349 adult patients with a mean age of 53 years. Among the patients, 52.7% of them identified as female and 47.3% of them identified as male; and 57.0%, 18.9%, 5.4%, 4.9%, and 10.9% of them identified as White, Black, Asian, Hispanic, and other or unknown, respecitvely. The study was conducted between January 1, 2020, and March 9, 2021 at two large academic medical centers in the northeastern United States.

The researchers assigned scores to the study participants based on factors such as initial serum lipase level, number of prior acute pancreatitis episodes, abdominal pain characteristics, and medical history. A final diagnosis of acute pancreatitis was established by expert review of hospitalization records.

The researchers found that 45.0% (n = 157) of the patients had a final diagnosis of acute pancreatitis. The most common non–acute pancreatitis diagnoses were nausea, vomiting, and gastroenteritis (18.2%); kidney failure (12.5%); and alcohol intoxication (8.9%).

The model demonstrated effective discriminatory accuracy with an area under the receiver operating characteristics curve of 0.91. At a diagnostic cutoff of at least 6 points, the model achieved a sensitivity of 81.5%, specificity of 85.9%, positive predictive value of 82.6%, and negative predictive value of 85.1% for diagnosing acute pancreatitis. The model's performance was also assessed at different diagnostic thresholds. At a more conservative cutoff of at least 8 points, specificity (96.4%) and positive predictive value (92.7%) were higher, whereas sensitivity (56.7%) and negative predictive value (73.1%) were lower.

Early abdominal imaging—computed tomography or magnetic resonance imaging within 24 hours—was performed more often in the participants predicted to have acute pancreatitis based on the model but rarely revealed alternative diagnoses. Among those with scores of at least 8 points, 98.6% of early imaging studies revealed acute pancreatitis or were nondiagnostic. The median time from the return of serum lipase results to early imaging was 86 minutes, and 11.9% of the participants underwent imaging prior to receiving their serum lipase results.

The researchers indicated that while most patients with acute pancreatitis fulfill diagnostic criteria at presentation, early imaging is often used for confirmation—despite guidelines stating it may not be necessary. Serious alternative diagnoses, such as bowel obstruction and perforation, were uncommon in the study population (4.6%). Applying the model with a cutoff of at least 6 points would have misdiagnosed only one participant (0.3%) with a serious alternative diagnosis.

Conflict of interest disclosures can be found in the study.